The effect of Fe oxidizing bacteria on F

Chemical Geology 180 Ž2001. 99–115
www.elsevier.comrlocaterchemgeo

The effect of Fe-oxidizing bacteria on Fe-silicate
mineral dissolution
Cara M. Santelli a,1, Susan A. Welch a,) , Henry R. Westrich b, Jillian F. Banfield a
a

Department of Geology and Geophysics, UniÕersity of Wisconsin-Madison, 1215 W. Dayton St., Madison WI 53706 1692, USA
b
Geochemistry Department, Sandia National Laboratories, Albuquerque NM 87185-0750, USA

Abstract
Acidithiobacillus ferrooxidans are commonly present in acid mine drainage ŽAMD.. A. ferrooxidans derive metabolic
energy from oxidation of Fe 2q present in natural acid solutions and also may be able to utilize Fe 2q released by dissolution
of silicate minerals during acid neutralization reactions. Natural and synthetic fayalites were reacted in solutions with initial
pH values of 2.0, 3.0 and 4.0 in the presence of A. ferrooxidans and in abiotic solutions in order to determine whether these
chemolithotrophic bacteria can be sustained by acid-promoted fayalite dissolution and to measure the impact of their
metabolism on acid neutralization rates. The production of almost the maximum Fe 3q from the available Fe in solution in
microbial experiments Žcompared to no production of Fe 3q in abiotic controls. confirms A. ferrooxidans metabolism.
Furthermore, cell division was detected and the total cell numbers increased over the duration of experiments. Thus, over the

pH range 2–4, fayalite dissolution can sustain growth of A. ferrooxidans. However, ferric iron released by A. ferrooxidans
metabolism dramatically inhibited dissolution rates by 50–98% compared to the abiotic controls.
Two sets of abiotic experiments were conducted to determine why microbial iron oxidation suppressed fayalite
dissolution. Firstly, fayalite was dissolved at pH 2 in fully oxygenated and anoxic solutions. No significant difference was
observed between rates in these experiments, as expected, due to extremely slow inorganic ferrous iron oxidation rates at pH
2. Experiments were also carried out to determine the effects of the concentrations of Fe 2q, Mg 2q and Fe 3q on fayalite
dissolution. Neither Fe 2q nor Mg 2q had an effect on the dissolution reaction. However, Fe 3q, in the solution, inhibited both
silica and iron release in the control, very similar to the biologically mediated fayalite dissolution reaction. Because ferric
iron produced in microbial experiments was partitioned into nanocrystalline goethite Žwith very low Si. that was loosely
associated with fayalite surfaces or coated the A. ferrooxidans cells, the decreased rates of accumulation of Fe and Si in
solution cannot be attributed to diffusion inhibition by goethite or to precipitation of Fe–Si-rich minerals. The magnitude of
the effect of Fe 3q addition Žor enzymatic iron oxidation. on fayalite dissolution rates, especially at low extents of fayalite
reaction, is most consistent with suppression of dissolution by interaction between Fe 3q and surface sites. These results
suggest that microorganisms can significantly reduce the rate at which silicate hydrolysis reactions can neutralize acidic
solutions in the environment. q 2001 Elsevier Science B.V. All rights reserved.
Keywords: Fe-oxidizing bacteria; Fe-silicate mineral dissolution; Thiobacillus ferrooxidans

)

Corresponding author. Fax: q1-608-262-0693.

E-mail address: swelch@geology.wisc.edu ŽS.A. Welch..
1
Current address: Woods Hole Oceanographic Institute, Woods
Hole MA, USA.

1. Introduction
Microbes utilize Fe for a variety of purposes and
can greatly influence biogeochemical cycling of Fe.

0009-2541r01r$ - see front matter q 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 0 0 9 - 2 5 4 1 Ž 0 1 . 0 0 3 0 8 - 4

100

C.M. Santelli et al.r Chemical Geology 180 (2001) 99–115

In anoxic environments, microbes can use ferric iron
oxyhydroxide phases as electron acceptors for respiration of organic compounds Žsee Ehrlich, 1996 for
review., thus, greatly increase rates of Fe release to
solution. Under oxidizing conditions, Fe-oxidizing

Bacteria and Archaea can utilize reduced Fe as an
electron donor in energy generation. This metabolism
is generally thought to predominate at low pH, where
rates of inorganic iron oxidation are slow. However,
a growing body of evidence indicates that these
organisms play a significant role in Fe cycling and
CO 2 fixation in more neutral pH solutions. For
example, chemolithotrophic iron oxidizers have been
detected in soils and ground water in microenvironments where pO 2 is low or at oxicranoxic interfaces ŽEmerson and Moyer, 1997; Emerson et al.,
1999.. Nitrate-reducing lithotrophic iron oxidizers
have also been detected in anoxic environments Že.g.,
Straub et al., 2001.. Observations interpreted to suggest the activity of Fe-oxidizing microbes associated
with oceanic basalts have been reported ŽThorseth et
al., 1995; Fisk et al., 1998.. In the case of near-neutral pH systems, iron-oxidizing microbes are usually
detected in groundwater mixing and discharge zones,
and the connection between reactions releasing reduced iron and microbial populations is unclear.
Specifically, the link between dissolution kinetics of
iron silicates and growth of Fe-oxidizing microbes
has received little attention. This is true both at low
pH and in near-neutral pH solutions associated with

weathered rocks.
Based on thermodynamic calculations, Jakosky
and Shock Ž1998. predicted that approximately 2 g
of microorganisms could be supported by the alteration of 1 mol of fayalite. However, the thermodynamic argument fails to consider whether the rate of
the mineral alteration reaction is sufficient to sustain
a community of Fe-oxidizing microorganisms. Recently, direct experimental evidence confirmed that
populations of Fe-oxidizing can be sustained solely
by fayalite dissolution at neutral pH ŽWelch et al.,
submitted..
Although Fe oxidizing chemolithotrophic microorganisms may have limited importance in silicate mineral alteration now compared to the effect of
heterotrophic organisms and plants, it is likely that
they have been very important in earlier geologic
periods. Phylogenetic analyses of small subunit ribo-

somal RNA sequences indicate that earliest organisms for which we have genetic signatures were
autotrophs, organisms capable of converting CO 2 to
organic carbon ŽBarns and Nierzwicki-Bauer, 1997..
Furthermore, metabolic gene sequence analyses suggest that many autotrophic pathways evolved billions
of years ago ŽBult et al., 1996.. Although the relative
importance of inorganic energy sources has changed

over geologic time with the evolution of higher life
forms, energy associated with reduced metals Žespecially Fe. in silicate minerals may have been and
remain quite significant.
The experiments reported here had two purposes.
Firstly, we aimed to determine whether rates of
dissolution of fayalite ŽFe 2 SiO4 . at pH 2–4 are rapid
enough and inorganic ferrous iron oxidation reactions slow enough, that significant populations of
Fe-oxidizing microorganisms can be sustained. Secondly, we measured the impact of enzymatic iron
oxidation on the kinetics of Fe-silicate mineral
weathering reactions. Our results have relevance both
to analyses of potential metabolic pathways for subsurface biospheres Žhere and on Mars. and to models
that predict acid neutralization rates Žacid mine
drainage, acid rain..

2. Experimental methods
2.1. Minerals
Natural fayalite was obtained from the American
Museum of Natural History ŽAMNH 10928. and
from the Smithsonian ŽR 3516.. Sample R 3516 is
from Cheyenne County, CO. Few millimeter-wide

crystal aggregates were crushed, sieved, washed and
characterized by X-ray diffraction and high-resolution transmission electron microscopy ŽHRTEM. and
SEM. The 25–75-mm size fraction was used for the
experiments. Synthetic fayalite was prepared by slow
cooling of a mixture of iron oxalate and quartz from
12758C to 11758C under a COrCO 2 atmosphere
Ž f O 2 –10y1 0 .. Synthetic fayalite had a particle size
of 25–75 mm and a surface area of 0.125 m2rg and
was used as received, without any sample pretreatment.

C.M. Santelli et al.r Chemical Geology 180 (2001) 99–115

2.2. Microbes
A lithotrophic iron oxidizing bacterium, Thiobacillus ferrooxidans, was used in the biotic experiments. A. ferrooxidans was cultured in a 9-K media
ŽSilverman and Lundgren, 1959 as cited in Ehrlich,
1996. on an orbital shaker at 258C at 130 rpm for
several weeks. The 9-K media consist of 3.0 g
ŽNH 4 . 2 SO4 , 0.1 g KCl, 0.5 g K 2 HPO4 , 0.5 g
MgSO4 P 7H 2 O, 0.01 g CaŽNO 3 . 2 , 44.22 g FeSO4 P
7H 2 O and 1-l autoclaved, distilled water. The bacteria were separated from their culture media by a

series of filtrations. The culture media with bacteria
was initially filtered using a paper filter Žapproximately 10-mm pore size. to separate out the iron
hydroxide precipitates from the media and bacteria.
The filtrate was collected and then, filtered again
using an autoclaved 0.22-mm polycarbonate filter to
collect the bacterial cells. This filter was flushed
with several milliliters of distilled water to remove
residual dissolved Fe from the bacterial cells. The
filter with bacterial cells was placed in a polycarbonate centrifuge tube with a few milliliters of sterile
nanopure water. The sample was then vortexed and
sonicated to remove cells from the filter. This solution was then examined by light and epifluorescence
microscopy for bacterial cells and for iron precipitates. Trace quantities of iron precipitates were detected in the cell slurry. The cell solution was stored
at 48C until the bacteria could be added to the
experiments, usually within several hours. All media
and solids were cleaned and autoclaved to minimize
possible contamination by other microorganisms. A.
ferrooxidans was an ideal organism to use in the
experiments because it can be grown in extremely
simple media Žsee below. and because its possible
growth substrates are limited to reduced iron and

sulfur Žit cannot utilize organic molecules..
2.3. Biological experiments
Fayalite dissolution experiments were carried out
in batch reactors. One hundred milligrams of fayalite
was added to 50 ml of solution in acid-washed
polycarbonate flasks. The solutions used for all biological experiments consisted of dilute H 2 SO4 , 1
mM NaHCO 3 , 1 mM ŽNH 4 . 2 SO4 , 100 mM K 2 HPO4

101

and filtered nanopure water. This media was chosen
since it would provide trace inorganic nutrients required by the lithotrophic microorganisms without
severely complicating the solution chemistry for the
dissolution experiments. Initial solution pH was adjusted by changing the molarity of sulfuric acid,
either 1 or 10 mM and then, titrated with either
dilute HCl or NaOH before experiments.
The experiments were done at an initial pH of 2,
3 and 4. Each set of experiments consisted of three
different treatments: an abiotic treatment with fayalite and solution; a biological treatment with approximately 2–3 ml of the cell solution, fayalite and
media and a fayalite-free control, which consisted of

2–3 ml of the cell solution in 50 ml of media. There
were approximately 10 7 cellsrml in the bulk solution in the biological and fayalite-free treatment. Cell
numbers remained relatively constant throughout the
course of the experiments. The fayalite-free control
was done to determine the amount of Fe and Si
associated with the bacterial cells. There were three
replicates for all abiotic and the biological experiments, except for the fayalite-free treatment Žwhich
was not replicated.. Flasks were placed on an orbital
shaker at 120 rpm to keep solutions aerated and well
mixed during the experiments. Five-milliliter aliquots
was taken periodically from each bottle for about 2
weeks using a sterile syringe. These aliquots were
filtered with a 0.22-mm acrodisc syringe filter to
remove the bacteria and mineral particles. Five
milliliters of fresh solution was then added to the
flasks to keep the volume constant. All experiments
were conducted at room temperature.
Experiments were performed at pH 2 using dead
cell controls to determine whether they influenced
rates of Fe and Si accumulation in solution. Cells

were AdeenergizedB Žinactivated, but not dead. using
1 mM Na 3 N Žsodium azide from Fisher Chemical.
ŽMera et al., 1992.. For this set of experiments, there
were five treatments, three of which are identical
to previous experiments: abiotic, live cells and fayalite-free. Two sets of flasks contained 1 mM Na 3 N:
an abioticq azide treatment to determine if Na 3 N
affected the dissolution reaction and cells q azide
treatment to determine the effect of dead Žor de-energized. cells on fayalite dissolution. The experiments were carried similarly to the previous ones,
except fresh solution was not added after sampling.

102

C.M. Santelli et al.r Chemical Geology 180 (2001) 99–115

2.4. Abiotic experiments
Since A. ferrooxidans had a significant effect on
the fayalite dissolution reaction, three sets of abiotic
experiments were conducted to distinguish the effect
of solution redox chemistry on fayalite dissolution
from effects due to reaction products. All abiotic

redox experiments were carried out at an initial
solution pH of approximately 2, under conditions
very similar to the abiotic controls in the previous
experiments.
One set of experiments tested the effect of dissolved oxygen on fayalite dissolution. The solution
was the same used in inorganic controls for the pH 2
biological fayalite dissolution experiments. One hundred microliters of solution and 100 mg of natural
fayalite were added to each reactor. In the oxic
treatment, oxygen was bubbled through the solution
for approximately 30 min, initially, and then, for 10
min everyday after a sample was taken. Similarly,
nitrogen was bubbled through the solution in order to
achieve essentially anoxic conditions. Bottles were
not constantly agitated but were swirled gently by
hand before samples were taken. A 5-ml aliquot was
taken periodically from each bottle over 10 days.
Samples were filtered with a 0.22-mm acrodisc filter.
Experiments were not replenished with fresh media
after sampling in order to minimize gas exchange.
A second set of abiotic fayalite dissolution experiments was carried out to determine the effect
of solution iron speciation on fayalite dissolution.
The Fe 2q vs. Fe 3q experiments consisted of four
different treatments with two replicates each: 10 mM
H 2 SO4 Žas a control., 10 mM H 2 SO4 plus 1 mM
Mg 2q, 10 mM H 2 SO4 plus Fe 2q, and 10 mM
H 2 SO4 plus Fe 3q. One hundred microliters of solution and 100 mg of synthetic fayalite were added to
polycarbonate bottles that were placed on a shaker
table for 8 days. Samples were taken periodically
using a syringe and filtered with a 0.22-mm filter.
A third set of abiotic fayalite dissolution experiments was performed to determine the effect the
varying amounts of Fe 3q in solution had on fayalite
dissolution. Fayalite was reacted in media containing
100 mM Fe 3q, 500 mM Fe 3q, 1 mM Fe 3q and 5
mM Fe 3q. The duplicated batch experiments consisted of 100 mg of fayalite reacted in 100 ml of
solution in polycarbonate flasks. Samples were col-

lected periodically for 10 days using a syringe and
filtered with a 0.22-mm filter. Fresh media was not
added after sampling.
2.5. Analytical
Analysis for Fe 2q and total Fe was done using the
ferrozine method ŽStookey, 1970.. Absorbance was
measured using a Perkin-Elmer UVrVIS spectrophotometer. Dissolved Si was measured using a
Technicon AutoAnalyzer II with the molybdate blue
method. Solution pH was determined with a Denver
Instruments pH meter that had been calibrated using
NBS buffers.
At the end of the dissolution experiments, a fraction of the mineral sample was removed from the
reactor and mounted on a SEM stub with carbon
tape. Mineral samples were gently rinsed and allowed to air dry. Samples were sputter coated with
either platinum or gold and examined with a LEO
982 high-resolution scanning electron microscope.
For some experiments, mineral samples were prepared for TEM analysis. Mineral grains were gently
crushed using a mortar and pestle, suspended in
deionized water and placed on a formvar-coated
copper grid. Samples were examined using a Philips
CM200 ultratwin transmission electron microscope
operated at 200 kV.
3. Results
The net concentrations of iron and silica released
were measured in both abiotic and biologic experiments. Results depended on solution pH and on the
presence or absence of microbial populations. Total
cell numbers in solution and on mineral surfaces
were estimated by direct counts using epifluorescence microscopy
3.1. pH 2 experiments
Fayalite dissolution experiments were conducted
with both natural ŽFig. 1. and synthetic ŽFig. 2.
fayalite. In the abiotic treatments, results were similar for both experiments, though concentrations were
approximately 40% lower for the synthetic vs. natural fayalite. Fe and Si release to solution was rapid in
the first few day of the experiment and then, leveled
off or slightly decreased with time Ždue to the addition of fresh media.. This ‘parabolic’ dissolution rate

C.M. Santelli et al.r Chemical Geology 180 (2001) 99–115

103

Fig. 1. Ža. Total Žopen symbols. and reduced Fe Žfilled symbols. concentration for natural fayalite dissolution at initial pH 2 for the controls
ŽB., microbial Žv . and fayalite-free Ž'. treatments. Žb. Si concentration from dissolving natural fayalite at initial pH 2 for the controls
ŽB., microbial Žv . and fayalite-free Ž'. treatments.

largely reflected the rapid consumption of protons in
the batch reactor. Bulk solution pH increased rapidly
from approximately 2 to 3.6 as the mineral dissolved. The initial Fe release rate calculated from a
linear fit of the Fe concentration curve over time
Žbefore Fe concentration becomes relatively constant. is ; 1.3–1.6 = 10y1 1 mol Fe 2qrcm2 s. Essentially, all of the iron in solution in the abiotic
experiments was as Fe 2q. The Si release curve is
very similar to Fe, though Si concentrations are
approximately half those of Fe, reflecting net stoichiometric dissolution ŽFigs. 1b and 2b.. In the
abiotic experiments, approximately 30% to 50% of
the fayalite had reacted by the end of the experiment,
thus, greatly limiting any contribution to the results
from surface layers modified by sample preparation.

In the biological experiments, both Si and Fe
release to solution was considerably lower than in
the abiotic controls. In the experiments with natural
fayalite, final concentrations were about a factor of
five lower than the abiotic controls. However, for the
experiments with synthetic fayalite, total concentrations were approximately 25 to 50 times less than the
abiotic experiments. The release of both Fe and Si
was nonlinear with time, but did not follow the same
rapid initial dissolution observed in the abiotic controls. In both biotic experiments, most of the measurable dissolved Fe is oxidized ŽFe 3q . as opposed to
the corresponding abiotic controls where Fe 3q was
undetectable. The net total FerSi release to solution
in the biotic natural fayalite experiments was approximately stoichiometric. Si was preferentially released

104

C.M. Santelli et al.r Chemical Geology 180 (2001) 99–115

Fig. 2. Ža. Total Žopen symbols. and reduced Fe Žfilled symbols. concentration for synthetic fayalite dissolution at initial pH 2 for the
controls ŽB., microbial Žv . and fayalite-free Ž'. treatments. Žb. Si concentration from dissolving synthetic fayalite at initial pH 2 for the
controls ŽB., microbial Žv . and fayalite-free Ž'. treatments.

compared to Fe in the synthetic fayalite experiments.
In the biotic experiments, solution pH increased from
2 to approximately 2.3. Even though the fayalite is in
acidic solution Žmaximum pH ; 2.3., it is apparently unreactive compared to the fayalite in the
abiotic controls.
The Fe and Si released in the fayalite-free control
was either undetectable or negligible compared to all
mineral dissolution experiments ŽFigs. 1 and 2..
Since live cultures of A. ferrooxidans appeared to
significantly inhibit fayalite dissolution under acidic
conditions, the experiments were repeated to determine if the inhibition was the result of cell metabolic
processes, or simply the result of adsorption of material to cell surfaces. The results of the control, biotic
and fayalite-free treatments were nearly identical to
the previous experiment done at pH 2 with synthetic

fayalite ŽFigs. 2 and 3.. Sodium azide had minimal
effect on the reaction. Total and reduced Fe released
from fayalite in the abiotic azide experiments was
only approximately 10% lower than the abiotic control, and Si release was indistinguishable from the
control. Similarly, fayalite dissolution in solutions
containing azide and dead cells were nearly identical
to the abiotic azide treatments. These results indicate
that dead cells had no significant effect on the
dissolution reaction under these experimental conditions.
Solution saturation state with respect to several
Fe and Si containing phases was calculated using
PHREEQC ŽParkhurst, 1995.. Within the first 2 days
of the experiments, solutions became supersaturated
by up to several orders of magnitude with respect to
goethite, hematite, quartz and chalcedony in both the

C.M. Santelli et al.r Chemical Geology 180 (2001) 99–115

105

Fig. 3. Ža. Reduced, Žb. total and Žc. Si concentrations from dissolving fayalite in dead cell experiments. Treatments are abiotic ŽB., azide
addition ŽI., dead cells with azide Ž`., biotic Žlive cells. Žv . and fayalite-free Ž'.. The abiotic, biotic, and fayalite-free treatments are
nearly identical to those in Fig. 2.

C.M. Santelli et al.r Chemical Geology 180 (2001) 99–115

106

Table 1
Saturation index ŽSI. with respect to several possible secondary
phases for conditions at the end of the dissolution experiments. SI
calculated using PHREEQC ŽParkhurst, 1995.

Biological pH 2
Biological pH 3
Biological pH 4
Abiotic pH 2
Abiotic pH 3
Abiotic pH 4

Goethite

Hematite

Quartz

SiO 2Ža .

1.41
2.76
3.67
3.00
2.09
–

4.83
7.53
9.35
8.00
6.18
–

0.18
y1.12
y1.41
1.6
y0.90
y0.98

y1.09
y2.38
y2.68
0.33
y2.17
y2.24

abiotic and biologic samples of the natural and synthetic fayalite experiments. Saturation index for conditions at the ends of the experiments are given in
Table 1.

The solution chemical data, in terms of net total
Fe and Si released to solution, indicates that approximately 2% to 40% of the fayalite had reacted over
the course of the experiments. Fig. 4 shows typical
high-resolution SEM images of natural fayalite reacted at pH 2 for 8 days. Although both sets of
minerals show evidence of dissolution, it is evident
from the SEM images that minerals in the abiotic
experiments are much more extensively reactive than
minerals in the biological experiments. Fig. 4a shows
a very extensively reacted natural fayalite grain from
the abiotic experiments. Deep etch channels and pits
formed on the mineral surfaces. Although these etch
channels are somewhat irregular, their orientation is
generally normal to c ) . The extent and pattern of
etching varies substantially for different surfaces.
The Ž001. surface appears to be relatively unreacted.

Fig. 4. Scanning electron microscope images of natural fayalite reacted in abiotic and biological experiments at initial pH 2. Ža, b. Fayalite
reacted in abiotic experiments. The surfaces are characterized by very extensive etching and deep holes and channels forming normal to c ) .
Žc. Fayalite reacted in with A. ferrooxidans. Etch texture is characterized by narrow planar etch channels normal to c ) . The micron-sized
ovals on the mineral surface are A. ferrooxidans cells. Žd. Higher magnification image of cells on the fayalite surface. Cells are surrounded
by extracellular polymers and iron oxyhydroxide precipitates. Note the dividing cell and flagella.

C.M. Santelli et al.r Chemical Geology 180 (2001) 99–115

Fig. 4b shows another very rough etched surface
with holes and channels forming parallel to Ž001.. In
the biotic experiments, the etching is dominated by
semi-periodic parallel channels formed parallel to
Ž001. ŽFig. 4c.. Based on SEM observations, these
etches appear to be typically 10s to 100s nm wide
and spaced approximately 1 mm apart. The micronsized ovals are bacteria on the fayalite surface. Fig.
4d is a higher magnification image of the mineral
surface showing dividing cells.
It was difficult to obtain accurate cell counts.
Data indicate that the total number of cells in suspension increased slightly Ž- a factor of 2. throughout the experiment. However, cell counts do not

107

include attached cells, as it was generally impossible
to detect them using epifluorescence microscopy.
Based on estimates from the SEM observations and
total cell counts, approximately 10–50% of the cells
are attached to mineral surfaces.
3.2. TEM and AEM characterization of reaction
products
HRTEM images and energy-dispersive X-ray
analyses Žanalytical electron microscopy; AEM. show
that the nanocrystalline Žfew nanometer in diameter.
particles of iron oxyhydroxides are produced in biological dissolution experiments. Interplanar spacings
in the product indicate a mixture of ferrihydrite and

Fig. 5. pH 3 Ža. Total Žopen symbols. and reduced Fe Žfilled symbols. concentration for natural fayalite dissolution at initial pH 3 for the
controls ŽB., microbial Žv . and fayalite-free Ž'. treatments. Žb. Si concentration from dissolving natural fayalite at initial pH 3 for the
controls ŽB., microbial Žv . and fayalite-free Ž'. treatments.

108

C.M. Santelli et al.r Chemical Geology 180 (2001) 99–115

goethite. Little Si was detected in secondary minerals. Particles were typically flocculated into aggregates and sometimes adhered to cells. They do not
form tightly adhering layers on altered fayalite surfaces.
3.3. pH 3 experiments
At pH 3, iron and silica release into solution is
about three orders of magnitude lower than for the
experiment at pH 2 with natural fayalite. This corresponds to approximately 0.1% of the solid material
reacting during the course of the experiments. Fig. 5
shows Fe and Si concentrations plotted against time.
In the abiotic treatment, most of the iron released
into solution is Fe 2q, whereas, in the biotic experi-

ments with A. ferrooxidans, most of the iron in
solution was Fe 3q, very similar to what was observed in the more acidic experiments. The dissolution reaction was not stoichiometric in both the
biotic and abiotic experiments. Si accumulates in
solution preferentially compared to Fe. The total
dissolved FerSi ratio was 0.5 Ža 4 = concentration
of Si compared to Fe, relative to fayalite. in the
biotic experiments compared to 1.0 in the abiotic
experiment. Fayalite dissolution rate at pH 3, based
on Si release, is 6.2 = 10y1 5 molrcm2 s in the abiotic experiment. Overall, the presence of microorganisms inhibited the release of silica by approximately a factor of two compared to the abiotic
experiments. The fayalite-free control showed relatively little dissolved iron or silica in solution.

Fig. 6. Ža. Total Žopen symbols. and reduced Fe Žfilled symbols. concentration for natural fayalite dissolution at initial pH 4 for the controls
ŽB., microbial Žv . and fayalite-free Ž'. treatments. Iron concentration in the biological and fayalite-free treatments are at the detection
limit for the method. Žb. Si concentration from dissolving natural fayalite at initial pH 2 for the controls ŽB., microbial Žv . and
fayalite-free Ž'. treatments.

C.M. Santelli et al.r Chemical Geology 180 (2001) 99–115

109

Fig. 7. Reduced and total Fe is solution from synthetic fayalite dissolution under oxic ŽB. and anoxic Žv . conditions.

Solution pH remained relatively constant with
time for both the abiotic and biologic experiments,
increasing from approximately 3.0 to 3.1 throughout
the course of the experiment. Solution saturation
state with respect to several possible Fe and Si
containing phases was calculated using PHREEQ.
All experiments are supersaturated with respect to
goethite and hematite by the end of the experiment
ŽTable 1..
3.4. pH 4 experiments
At pH 4, iron and silica release from synthetic
fayalite dissolution was approximately three orders

of magnitude lower than for the experiments at pH 2.
The abiotic fayalite dissolution rate, based on a
linear fit of the Si release curve, is 4.6 = 10y1 5
molrcm2 s. The total Fe and Si released into solution
was significantly greater in the abiotic controls compared to the biotic experiments ŽFig. 6.. However, in
contrast with the pH 2 and 3 experiments, total Fe in
the pH 4 experiments was dominated by Fe 2q in
both the abiotic and biologic experiments. However,
Fe concentrations are extremely low in the biological
experiments, very close to the detection limit for the
analytical method. The reaction was not stoichiometric. Silica accumulates in solution preferentially over
iron, especially for the biologic experiments Žwhere
iron is barely detectable.. The fayalite-free control

Fig. 8. Total iron release from fayalite dissolution at initial pH 2 in 1 mM solutions of Fe 2q ŽB., Mg 2q Žl., Fe 3q Ž'., and a control Žv ..

110

C.M. Santelli et al.r Chemical Geology 180 (2001) 99–115

showed no detectable iron or silica in solution. Solution pH remained relatively constant increasing from
approximately 4.0 to 4.3 for both the biologic and

abiotic samples. Solutions were supersaturated with
respect to goethite and hematite in the biologic experiments ŽTable 1..

Fig. 9. Ža. Reduced iron, Žb. total iron and Žc. Si from fayalite dissolution at initial pH 2 in a control ŽB., and with 100 mM Žl., 500 mM
Žv ., 1 mM Ž'. and 5 mM Ž . Fe 3q added to solution.

C.M. Santelli et al.r Chemical Geology 180 (2001) 99–115

4. Effect of solution redox conditions on fayalite
dissolution
4.1. Anoxic Õs. oxic experiments
Abiotic experiments were carried out at pH 2 to
determine if the dissolved oxygen concentration affected the fayalite dissolution rate. Fig. 7 shows iron
released into solution for both oxygenated and anoxic
Žor low oxygen. conditions. Results were very similar to the natural fayalite abiotic experiments at pH
2. Dissolved Fe increased with time for the first few
days of the experiment and then, the Fe concentration was approximately constant. The total Fe was
essentially all as Fe 2q, even in the solutions that
were supersaturated with respect to O 2 . There was
no significant difference between Fe release rate, and
Fe concentration in the two experiments, indicating
that under these experimental conditions, dissolved
O 2 has a negligible effect on the fayalite dissolution
reaction.
4.2. Fe 2 q and Mg 2 q experiments
Additional experiments were carried out at pH 2
to determine the effect of the concentration of other
ions, Fe 2q and Mg 2q, on natural fayalite dissolution.
Fig. 8 shows average total iron released into solution
over the course of the experiments. Results are nearly
identical to the previous abiotic experiments. All
dissolved iron released is Fe 2q. Neither the added
Fe 2q nor Mg 2q in solution had a significant effect
on fayalite dissolution compared to the sulfuric acid
control.

111

concentrations released into solution in the control
experiments were very similar to the abiotic control
experiments done at pH 2 with synthetic fayalite
ŽFigs. 2 and 3.. The reaction was stoichiometric. All
the dissolved iron was reduced and approximately,
40% of the fayalite dissolved by the end of the
experiment. Addition of 100 mM Fe 3q to solution
decreased Fe and Si release by almost 1r3 compared
to the controls. However, the reaction was not stoichiometric. The dissolved total Fe was approximately 1000 mM lower than the concentration predicted from Si for the stoichiometric dissolution of
fayalite. The lost Fe was approximately equivalent to
the precipitation of a goethite layer 150-nm thick
over the initial mineral surface.
The addition of 500 and 1000 mM Fe 3q to solution decreased net Fe and Si release from fayalite by
60% and 90% compared to the control. The reaction
was approximately stoichiometric, though the measured total Fe in solution was slightly lower than
added Fe 3q plus Fe 2q predicted from Si release
from fayalite dissolution. This difference between
measured and predicted iron would be equivalent to
a goethite layer that was only a few tens nanometers
thick over the initial mineral surface area.
Net Si release from fayalite in the experiments
with 5 mM Fe 3q is only 3% of Si in the control at
the end of the experiment. Approximately, 80% to
90% of the added iron precipitates during the experiment. This could correspond to a goethite layer
600–700-nm thick over the mineral surface.

5. Discussion

4.3. Fe 3 q addition experiments
5.1. Abiotic rates
Addition of 1 mM Fe 3q greatly inhibited iron
release from the fayalite. Total Fe released to solution from natural fayalite in the Fe 3q treatment was
approximately 30% of the control at the end of the
experiment ŽFig. 8.. This effect was comparable to
the apparent biological inhibition observed for fayalite dissolution for similar conditions ŽFig. 1..
Since both the biotic and abiotic experiments
appear to show a link between Fe 3q and inhibition
of the dissolution reaction, a series of experiments
were carried out with increasing amounts of Fe 3q Ž0,
0.1, 0.5, 1, 5 mM. in solution ŽFig. 9.. Fe and Si

In all of our abiotic batch dissolution experiments,
the release of Fe and Si to solution is nonlinear with
time. This nonlinear release of mineral components
to solution is a very typical feature of batch dissolution experiments and it reflects the changes in experimental conditions over time. The very rapid initial
dissolution rates are due primarily to proton consumption and preferential dissolution of more reactive material. The slower release rates reflect conditions where solution pH is higher, and solutions are
supersaturated with respect to several possible sec-

112

C.M. Santelli et al.r Chemical Geology 180 (2001) 99–115

ondary phases and chemical affinity for fayalite dissolution approaches zero. Rates also decrease with
time because 30–50% of the available material reacts within the first few days of the dissolution
experiments.
Abiotic fayalite dissolution rates can be calculated
from the initial stage of the reaction where Si and Fe
are rapidly released to solution and no precipitation
occurs. The abiotic dissolution rate of fayalite determined from the increase in Fe or Si in solution
ranged from approximately 1 = 10y1 1 molrcm2 s at
initial pH 2 to 5 = 10y1 5 molrcm2 s at initial pH 4.
These rates are comparable to other experimentally
determined dissolution rates for Fe-silicate minerals.
For example, Westrich et al. Ž1993. calculated a
fayalite dissolution rate of 6 = 10y1 1 molrcm2 s at
pH 2, slightly faster than the rates determined in this
study. In our pH 2 experiments, acidity decreased
with time as the mineral dissolved, therefore, the rate
is for an average pH of approximately 2.4. Wogelius
and Walther Ž1992. measured fayalite ŽFo 6 . dissolution rates in flow through reactors of 5 = 10y1 2 to
7 = 10y1 4 molrcm2 s from pH 2 to 5, rates very
similar to those we estimated from our batch experiments. Experimentally determined mineral dissolution rates typically vary by as much as on order of
magnitude depending on experimental conditions,
i.e., batch or flow through reactors, mineralrsolution
ratio, mineral composition, mineral preparation, pretreatment, etc., so our experimental rates are consistent with others. However, mineral dissolution experiments conducted, where all conditions are held
constant and only one parameter is varied, tend to be
very reproducible Žsee for example Figs. 2 and 3.,
and differences between experiments, either in terms
of rate of release of mineral components or concentrations in solution, can be attributed to the affect of
that variable Ži.e., changing solution pH, the presence
of microbes, dissolved O 2 content, etc...
5.2. Effect of T. ferrooxidans on fayalite dissolution
Fayalite dissolution experiments conducted in the
presence of iron-oxidizing microbes showed a suppression of both iron and silica release into solution
compared to abiotic controls. SEM images, which
show mineral grains from the abiotic experiments are

much more reacted than those from the biologic
experiments of the reacted minerals at pH 2, confirming the solution geochemical results. The apparent inhibition of the reaction could be due to several
factors.
A decrease in mineral dissolution rates over time
has been attributed to the formation of a diffusion
inhibiting leached layer or secondary precipitates on
the mineral surface ŽSchott and Berner, 1983, 1985;
Chou and Wollast, 1984..
The formation of an iron-rich product on the
fayalite surface is consistent with the solution chemistry, which shows a preferential accumulation of Si
compared to Fe in solution in the biological experiments. Furthermore, solutions are apparently supersaturated with respect to several possible secondary
Fe phases ŽTable 1.. In the Fe 3q addition experiments, the inhibition increases with increasing Fe 3q
added to solution. In the biotic experiments at pH 2,
total Fe concentration at the end of the dissolution
experiment is approximately 150 mM less than what
would have been predicted from Si concentration for
stoichiometric fayalite dissolution. If all this iron
precipitated as goethite, it would form a 15-nm-thick
layer over the fayalite surface Žassuming a fayalite
surface area of 1000 cm2rg and no precipitation in
solution or the reaction vessel.. However, there is no
evidence, at least in the biotic experiments, that the
goethite has formed a layer that could inhibit diffusion of ions to or from the mineral surface. TEM
Žalso see Welch and Banfield, in press. and SEM
images show no sign of an extensive leached layer or
precipitated layer on the mineral surface. In fact, the
particulate form of the goethite produced suggests
that this product does not significantly modify the
reactivity of the fayalite surface. Furthermore, surface reaction rates and therefore, dissolution rates,
are generally thought to be considerably slower than
diffusion rates. Thus, we conclude that a diffusioninhibiting layer composed of iron oxyhydroxide minerals does not explain the suppression of fayalite
dissolution rates in the presence of iron-oxidizing
microorganisms.
In the experiments at pH 3 and 4, there is an
approximately 50% inhibition of the fayalite dissolution rate compared to the abiotic experiments. It is
again unlikely that this is due to formation of a
diffusion inhibited barrier on the surfaces, especially

C.M. Santelli et al.r Chemical Geology 180 (2001) 99–115

considering the total amount of material reacted in
these experiments is not sufficient to produce a one
unit cell thick layer over the mineral surface.
Another explanation for the suppression of fayalite dissolution rates in the presence of microorganisms or Fe 3q is that Fe 3q absorbs to and chemically
passivates the mineral surface. The chemical passivation of the surfaces has been suggested for other
minerals. For example, Al 3q adsorption to feldspar
surfaces appears to substantially decrease their reactivity under some experimental conditions Že.g.,
Oelkers et al., 1994. by forming stable cross-links
between isolated Si tetrahedra. Alternatively, other
experimental studies have failed to show a significant inhibitory affect Že.g., Chen and Brantley, 1997,
2000.. Fe 3q, like Al 3q, should be able to adsorb to
mineral surfaces. The rates of solvent exchange
around the Fe 3q ion are about three orders of magnitude slower than rates surrounding Fe 2q, therefore,
the reactivity of the Fe 3q`O bond should be orders
of magnitude slower than the Fe 2q`O bond ŽCasey
et al., 1993; Casey and Westrich, 1992.. Consequently, fayalite surfaces that have adsorbed Fe 3q or
Fe ions that have been oxidized should be significantly less reactive than those with Fe 2q surface
sites. Furthermore, naturally weathered olivine is
highly reactive, except in regions containing planar
defects of laihunite, an olivine in which Fe 3q is
substituted primarily in the M2 octahedral sites and
M1 vacancies Ži.e., wx x ŽFe 2q . 2y3 x ŽFe 3q . 2 x SiO4 .
ŽBanfield et al., 1989, 1990, 1992; Casey et al.,
1993.. The lower reactivity of ferric iron-rich olivine
is consistent with a passivation mechanism due to
adsorbed ferric iron as proposed here.
5.3. Effect of solution redox chemistry on fayalite
dissolution
It is apparent from our experiments that solution
redox reactions can dramatically affect rates and
mechanisms of Fe-silicate mineral weathering. Our
experiments clearly demonstrate that Fe 3q in solution, either produced by A. ferrooxidans or added to
solution abiotically, inhibited fayalite dissolution under acidic conditions. The effect increased with increasing Fe 3q in solution. The presence of dissolved
oxygen, which can abiotically oxidize Fe 2q to Fe 3q,

113

had no detectable affect on the reaction. However,
under the experimental conditions, the abiotic oxidation rate of Fe 2q is extremely slow, approximately
0.1 mmolrlrday ŽNordstrom and Southam, 1997.,
therefore, Fe 3q was not detected during the duration
of the abiotic experiments. This example clearly
emphasizes that microorganisms can dramatically alter the rates and products of geochemical reactions
by ensuring that reactions occur under conditions
that would not be encountered in the absence of a
catalyst Žthe enzymes of the electron transport chain..
Others researchers have reported a decrease in
Fe-silicate reaction rates under oxic conditions due
to the oxidation of Fe 2q. For example, Wogelius and
Walther Ž1992. noted a decrease in fayalite dissolution rates over time in their flow through reactors.
They attributed this to the slow oxidation of reduced
iron and precipitation of ferric hydroxides on the
mineral surface. Siever and Woodford Ž1979. saw an
increase in dissolution of many Fe containing rocks
and minerals, fayalite, hypersthene, basalt and obsidian under slightly acidic anoxic conditions compared
to air saturated or O 2 saturated solutions. Total
amount of material released to solution was up to an
order of magnitude under anoxic conditions greater
than comparable experiments in equilibrium with air
or O 2 , presumably due to the precipitation of iron
phases on mineral surfaces. Alternatively, several
researchers reported an increase in dissolutionr
weathering rate of Fe silicates, due either to oxygen
or another oxidizing agent. Hoch et al. Ž1996. saw a
significantly higher flux of Si, Ca and Mg from
augite in their 1.5-ppm O 2 experiments at near-neutral pH compared to their 0.6-ppm O 2 experiments,
whereas, dissolved O 2 content had no apparent effect
on diopside dissolution. Diopside dissolution rates
were lower than augite rates. They attributed the
increase in reaction rate with increasing dissolved O 2
and faster rates in Fe-rich minerals to the oxidative
dissolution of Fe from the mineral. In a similar
study, White and Yee Ž1985. measured an enhancement of augite dissolution in the presence of dissolved O 2 and Fe 3q. They attributed this increase in
reaction rate to an electron transfer between Fe 3q in
solution and Fe 2q on the mineral surface. The resulting formation of a surface Fe 3q leads to a charge
imbalance and an increase in the flux of other ions to
solution.

114

C.M. Santelli et al.r Chemical Geology 180 (2001) 99–115

5.4. Microbial growth from Fe silicate minerals
If mineral dissolution rates and microbial iron
oxidation rates are known, then, its possible to estimate the numbers of iron oxidizing microorganisms
that could be sustained by iron oxidation. Based on
an iron oxidation rate of ; 6 = 10y1 8 mol Fe 2qr
cells ŽWelch et al., in preparation. and a fayalite
dissolution rate of ; 10y1 0 –10y11 molrcm2 s Žfrom
Westrich et al., 1993., we predict that approximately
10 8 –10 9 cells can be sustained by fayalite dissolution
at pH 2 in our experiments over the course of the
experiment. This is consistent with the numbers of
cells detected on our experiments for slightly slower
reaction rates.
5.5. Implications for rates of acid neutralization by
silicate mineral hydrolysis
The results of our experiments may be relevant to
a variety of natural processes, as well as to reactions
that occur in sites impacted by acid rain or acid
drainage. Even though bulk solution pH for most
natural environments is typically Anear neutralB, it is
not unreasonable that in microbial microenvironments, conditions could be significantly more acidic
ŽBarker et al., 1998.. Results of this study show that
under such conditions, iron silicate dissolution reactions could sustain significant microbial populations.
Our results do not rule out the ability of silicate
dissolution reactions to sustain microbial populations
at near-neutral pH, but it is probable that the number
of organisms would be lower due to the slower
Fe-silicate dissolution rate and the communities
would be confined to low oxygen environments
where inorganic iron oxidation rates are slow. Neutrophilic iron-oxidizing microbes are very difficult to
cultivate in the laboratory, so they were not used in
the current experiments. However, they may be common in natural microaerophilic environments ŽEmerson and Moyer 1997; Emerson et al., 1999. including environments expected at Fe-silicate surfaces
early in chemical weathering of rocks.
At near-neutral pH, the solubility of ferric iron is
so low that, once formed by enzymatic routes, it
precipitates very close to the oxidation site Že.g., on
the cell surface; see Banfield et al., 2000.. Under
these conditions, it is likely that ferric iron suppres-

sion of fayalite dissolution will not occur. Thus, we
predict that fayalite rates will not be as strongly
inhibited by iron oxidation in near-neutral pH solutions Žin contrast to the process at low pH, where
ferric iron has a significant residence time in solution..

6. Conclusion
The results of our experiments demonstrate that
fayalite dissolution at low pH is able to sustain an
active population of the iron-oxidizing bacterium T.
ferrooxidans. Metabolic activity is evidenced by the
production of almost the maximum concentration of
ferric iron in solution, persistence and increase in
size of populations over the duration of experiments,
production of precipitates in proximity to cells and
presence of dividing cells on the fayalite surface.
Our results also show that the fayalite dissolution
rate increases by three orders of magnitude between
pH 4 and 2 implying a much larger microbial population can be sustained at low pH. However, the
activity of the iron oxidizing bacterium suppresses
the rates of fayalite dissolution at low pH. The
results of abiotic, ferric iron-supplemented controls,
are attributed to oxidation of Fe 2q to Fe 3q under
conditions where ferric iron remains relatively soluble for long enough to interact with the fayalite
surface. These results are relevant to conditions encountered during neutralization of acid mine drainage
or when acid rain interacts with silicate rocks
Žolivine-group minerals being the most reactive of
the common silicate phases..

References
Banfield, J.F., Veblen, D.R., Jones, B.F., 1989. TEM of
laihunite-olivine intergrowths in oxidized olivine, and their
influence on weathering reactions. Geological Society of
America, 1989 annual meeting., 21 ŽEd. Anonymous., 240.
Geological Society of America ŽGSA..
Banfield, J.F., Veblen, D.R., Jones, B.F., 1990. Transmission
electron microscopy of subsolidus oxidation and weathering of
olivine. Contrib. Mineral. Petrol. 106, 110–123.
Banfield, J.F., Dyar, M.D., McGuire, A.V., 1992. The defect
microstructure of oxidized mantle olivine from dish hill, California. Am. Mineral. 77, 977–986.

C.M. Santelli et al.r Chemical Geology 180 (2001) 99–115
Banfield, J.F., Welch, S.A., Zhang, H., Ebert, T.T., Penn, R.L.,
2000. Crystal growth and microstructural evolution of FeOOH
biomineralization products. Science 289, 751–754.
Barker, W.W., Welch, S.A., Chu, S., Banfield, J.F., 1998. Experimental observations of the effects of bacteria on aluminosilicate weathering. Am. Mineral. 83, 1551–1563.
Barns, S.M., Nierzwicki-Bauer, D.H., 1997. Microbial diversity in
ocean, surface and subsurface environments. In: Banfield, J.F.,
Nealson, K.H. ŽEds.., Geomicrobiology: Interactions between
Microbes and Minerals. Mineral. Soc. Am. Rev. Mineral., vol.
35, pp. 35–79.
Bult, C.J., White, O., Olsen, G.J., Zhou, L.X., Fleischmann, R.D.,
Sutton, G.G., Blake, J.A., FitzGerald, L.M., Clayton, R.A.,
Gocayne, J.D., Kerlavage, A.R., Dougherty, B.A., Tomb, J.F.,
Adams, M.D., Reich, C.I., Overbeek, R., Kirkness, E.F.,
Weinstock, K.G., Merrick, J.M., Glodek, A., Scott, J.L., Geoghagen, N.S.M., Weidman, J.F., Fuhrmann, J.L., Nguyen, D.,
Utterback, T.R., Kelley, J.M., Peterson, J.D., Sadow, P.W.,
Hanna, M.C., Cotton, M.D., Roberts, K.M., Hurst, M.A.,
Kaine, B.P., Borodovsky, M., Klenk, H.P., Fraser, C.M.,
Smith, H.O., Woese, C.R., Venter, J.C., 1996. Complete
genome sequence of the methanogenic archaeon, Methanococcus jannaschii. Science 273, 1058–1073.
Casey, W.H., Westrich, H.R., 1992. Control of dissolution rates of
orthosilicate minerals by divalent metal–oxygen bonds. Nature
ŽLondon. 355, 157–159.
Casey, W.H., Banfield, J.F., Westrich, H.R., McLaughlin, L.,
1993. What do dissolution experiments tell us about natural
weathering. Chem. Geol. 105, 1–15.
Chen, Y., Brantley, S.L., 1997. Temperature